Mechanisms of Radiation Effects in the Natural Space Radiation Environment

111 Pages·1994·12.029 MB·English

Basic Mechanisms of Radiation Effects in the
Natural Space Radiation Environment
James R. Schwank
Sandia National Laboratories
Albuquerque, NM 87185-1083
DISCLAIMER
This report was preparedasanaccountofworksponsoredbyanagencyofthe UnitedStates
Government. Neither the United States Governmentnorany agencythereof, noranyof their
employees, makesanywarranty,expressorimplied,orassumesanylegalliabilityorrupoasi-
bility fortheaccuracy,completeness,orusefulnessof anyinformation,apparatus,product,or
processdisclosed,or represents that itsusewould notinfringeprivatelyowned rights.Refer-
encchereinto anyspecificcommercialproduct,process,or servicebytradename,trademark,
manufacturer, or otherwise doesnot necessarilyconstitute or implyits endorsement,rccom-
mendation, or favoringby the United States Governmentorany agency thereof.The views
and opinions of authors expressed herein do not necessarilystate or reflect those of the
United StatesGovernmentoranyagencythereof.
This work was supported by the U. S. Department ofEnergy through contract
number DE-AC04-94AL85000.
OlSTRtlBUTION OF THI6 DOCUMENT IS UNLIMITED 0 _
Basic Mechanisms of Radiation Effect_ in the
Natural Space Environment
James R. Schwank
Sandia National Laboratories
Radiation Technology and Assurance Department
1.0 Introduction
2.0 Natural Space Radiation Environment
2.1 Particles trapped by the earth's magnetic field
2.2 Cosmic rays
2.3 Radiation environment inside a spacecraft
2.4 Laboratory radiation sources
3.0 Interaction of Radiation with Materials
3.1 Ionization effects
3.2 Displacement effects
4.0 Total-Dose Effects--MOS Devices
4.1 Measurement techniques
4.2 Electron-hole yield
4.3 Hole transport
4.4 Oxide traps
4.5 Interface traps
4.6 Border traps
4.7 Device properties
4.8 Case studies
4.9 Special concerns for commercial devices
5.0 Total-Dose Effects Other Device Types
5.1 SOI devices
5.2 Nitrided oxide devices
6.0 Single-Event Phenomena
6.1 Mechanisms of charge collection
6.2 Hard errors
7.0 Summary
1.0 INTRODUCTION
Electronics in a satellite system can be degraded significantly by the natural space
radiation environment. A major goal of the radiation effects community has been to provide
devices that can function as intended in the harsh environment of space. This has required the
development of process techniques to fabricate radiation-hardened devices and the development
of reliable, cost-effective hardness assurance test procedures. To qualify a device for use in a
space system, one must rely on laboratory measurements typically at dose rates from 50 to 300
rad/s in which the radiation exposure may take only minutes to hours to complete. These
laboratory measurements must be correlated to a space environment in which the radiation
1I-2
exposuremaytakeplaceoveraperiodof manyyears. Tomakethesecorrelations,it isnecessary
to haveathoroughunderstandingof the mechanismsthat governthe radiationresponseof the
devices to be used. This is especiallytrue forsystems emploYingcommercial,non-radiation-
hardeneddevices where the marginbetweensystem requirementsanddevice capabilityis much
lower than for radiation-hardeneddevices. Thus,as commercial devices become increasingly
morepopular,the needforunderstandingradiation-responsemechanisms becomesincreasingly
moreimportant. Knowledgeof themechknismsof deviceradiationresponsehas alsoenabledthe
fabricationof radiation-hardeneddevices. Therefore,understandingthe basic mechanismsof
radiationeffects isof practicalimportancetothe system,design,andtechnologyengineer.
In this portion of the ShortCourse,the basic mechanismsof radiationeffects in the
naturalspace environmentare presented. The primarymannersin which the natural space
environmentcan cause degradationof electrical devices and systems are through total-dose
ionizing-radiationdamage,single-eventrelatedsoft andhard errors, anddisplacementdamage.
Of thesethree,Icover total-doseandsingle-eventeffects. Thegoal of thisportionof thecourse
is to providethe student with the basic knowledge requiredto understandthe mechanisms
underlying the development of hardness assurance test guidelines and hardened-process
technologies. Knowledge of the mechanisms will give the studentmoreconfidenceinapplying
hardnessassurancetest guidelinesfor space and otherapplications. This portion of the Short
Courseis also intendedto set the stage andprovidethe fundamentalsfor the remainderof the
ShortCourse. Althoughthe materialpresentedfocuses on device responseinthe naturalspace
environment,much of the materialpresentedis also applicableto the mechanismsof device
responseatshorttimes afterapulseof irradiation(e.g.,weaponapplication)anddevice response
formoderate-dose-rateexposures.
Webeginwithadescriptionof the naturalspaceradiationenvironment. This isthe first
step in determiningthe mechanismsgoverningdevice response. The mechanisms of device
response depend on the type, energy, and concentrations of particles present in the space
environment. The second step is to study the mannerin which the particles interact with
materials. Forinstance, protonscan cause total-ionizing-radiationdamage,single-eventupset,
and displacementdamage. On the other hand, electrons cause primarilytotal-dose ionizing
radiationdamageandhigh-energyions causeprimarilysingle-eventsoft andhard errors. Once
the manner in which radiationinteracts with materials is determined, the third step is to
determinethe mechanisms that govern the responsefor the device type of interest for the
particle(s)of interest. Ifocus inthisportionof theShortCourseon the mechanismsthatgovern
the total-dose response of MOS devices. MOS devices constitute a major portion of the
electronics of nearly all modem space systems. The material presentedcan be appliedto the
understandingof both commercialandradiation-hardeneddevice response. Knowledge of the
mechanismsthat governMOS device responsecan also be used to understandthe mechanisms
governinganumberof otherdevicetypes,includingSOIandSOSdevices andleakagecurrentin
advancedbipolarintegratedcircuits. Ipresentthe mechanismsof MOSdevice responseatshort
times following high-dose-rateirradiations. Althoughknowledge of the short time responseis
not important for characterizing low-dose-rate space irradiations, the short-time response
providesinsightintothe mechanismsgoverningradiationeffects inbothhigh-andlow-dose-rate
environments.Examples of case studieswhereknowledgeof the basicmechanismsof radiation
1I-3
effects hasied to technological improvements in device hardening and in hardness assurance test
methodology are presented. I next discuss the basic mechanisms of device response for two
device types that may see increased commercial use in the future: SOI and nitrided oxide devices.
Finally, ! cover the basic mechanisms of charge collection in silicon and GaAs devices leading to
single-event effects. Mechanisms are covered at the transistor level. The mechanisms for heavy-
ion induced single-event burnout and single-event gate _pture are also discussed.
2.0 NATURAL SPACE RADIA_ON ENVIRONMENT
The natural space environment can cause damage to electronic systems in a number of
ways. Itcontains high energy protons and electrons that can cause total-dose ionizing radiation-
induced damage. Protons can also cause displacement damage. Heavy ions and high-energy
protonscan upset system operation and sometimes cause permanentdamage to electronics. The
concentration and types of particles vary significantly with altitude and angle of inclination,
recent solar activity, and amount of spacecraft shielding. As such, it is nearly impossible to
define a ',typical" space environment. Particles present in the earth's natural space radiation
environment can be grouped into two general categories: 1) particles trapped by the earth's
magnetic field (primarilyelectrons and protons), and2) cosmic rays:heavy ions and high-energy
protons of galactic or solar origin. Inthis section, some of the general properties of the natural
space environment are presented.
2.1 ParticlesTrappedby the Earth'sMagnetic Field
The earth's magnetic field creates a geomagnetic cavity known as the magnetosphere [1].
The magnetic field lines trap low-energy charged particles. These trapped particles consist
primarily of electrons and protons, although some heavy ions are also trapped. The trapped
particles gyrate spirally around the magnetic field lines and are reflected back and forth between
the poles where the fields are confined. The motion of the trapped particles is illustrated in Fig. 1
[1]. As charged particles gyrate along the magnetic field lines, they also drift around the earth
with electrons drifting in an easterly direction and protons drifting in a westerly direction. The
motion of charged particles forms bands (or domains) of electrons and protons around the earth
and form the earth's radiation belts.
The boundaries of the domains at the equator are illustrated in Fig. 2 [1]. Distances are
specified in earth radii (one earth radius is equal to 6380 km) referenced to the center of the
earth, i.e., one earth radius isat the earth's surface. Because of the variation in the magnetic field
lines with latitude, the boundaries of the domains vary with latitude (angle of inclination). Most
satellites are operated in near-earth orbits at altitudes from slightly above 1 earth radius to 10
earth radii. Geosynchronous orbit (GEO) is at an altitude of approximately 35,800 km
corresponding to approximately 6.6 earth radii. The domains can be divided into five regions.
The trapped proton distribution exists primarily in regions one and two that extend from slightly
above 1earth radius to 3.8 earth radii. The distribution of proton flux as a function of energy and
radial distance is given in Fig. 3 [1]. [Flux is the rate at which particles impinge upon a unit
surface area. It is normally given in units of particles/cm2-s. The time integral of flux is the
fluence. Thus, fluence is equal to the total number of particles that impinge upon a unit surface
11-4
" *_,-,:__, "_,_ _ "_: -(!. _ _.. i' _. ... ••_ _.- -"
I |
I ;dltl__ _t1_._ "mFmOaPOINT V
Figure1: Motionoftrappedparticlesintheem'th'smagnetosphere.(AfterRef.1)
areafor agiventime intervalandit isnormallygiveninunitsof particles/cm2.]Trappedprotons
can have energiesas high as 500MeV [1]. Note that the altitude correspondingto the peak in
flux decreaseswith proton energy for any given energy. Protonswith energies greaterthan
10MeV primarilyoccupy regions one and two below 3.8 earth radii [1]. Typical spacecraft
shielding attenuates protons with energiesbelow 10MeV [2]. Thus, the predominantlylow-
energy trapped protons present above 3.8 earth radii are normally ineffective in producing
radiation-induceddamage. ',_'orproton energies greaterthan 30MeV, the highest proton flux
occurs at about 1.5 earth radii. P;otons originating from solar flares (discussed below) are
presentpredominantlyin regionsfour andfive (Fig.2) andextend from--5earthradiitobeyond
14earthradii.
Above the Atlantic Ocean off the coast of South Americathe geomagnetic sphere dips
towardthe earthcausing aregionof increasedprotonflux at relativelylow altitudes. Thisregion
iscalled the South Atlantic anomaly (SAA). In this region, the flux forprotons with energies
greaterthan 30MeV can be as muchas 104times higherthan in comparablealtitudesoverother
regions of the earth. At higher altitudes the magnetic sphere is moreuniform and the South
Atlanticanomalydisappears [3].
Electrons are presentpredominantlyin regionsoneto fourandextendupto 12earthradii
[1]. Theelectrondomain is dividedinto two zones, an inner zone extendingto abo_t 2.8 earth
radii and an outer zone extending from2.8 to 12earth radii. The outer zone electrons have
higherfluxes(~10times) andenergiesthanthe innerzone electrons. Forelectronswithenergies
greaterthan 1MeV,the peak in flux is locatedbetween 3 and4 earthradii[4]. The maximum
energyof trappedelectrons is approximately7MeV in the outer zone; whereas, the maximum
energy is less than 5MeV for electrons in the inner zone [1]. At these energies electron
interactionsareunimportantforsingle-eventeffects, butmustbe consideredindeterminingtotal-
doseeffects.
II-5
| |
-3.8
-2.8 -5.0 -12.0
Region 1 1213 : 4 ,, 5
I I
Solar Flare i I
I I
Trapped I I
Protons
Outer Zone
Electrons
Inner Zone
Electrons
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Earth Radii
Inner Zone Trapped
Electrons Protons
Outer Zone _:_'_'_"_S'_o::la'_r'_F'_la':r':e:_'_
Electrons ...._........_....;..:..;._.P...r_..o,..t.o..n..s
Figure 2: Boundaries of the domains for solar flare and trapped protons and outer andinner
zoneelectrons.(AfterRef. 1)
Fluxes of electrons and protons in particular orbits can be estimated from existing
models. Two models that provide reasonable estimates of the proton and electron fluxes as a
function of the satellite orbitareAP8 [5] forprotonsand AE8 [6] forelectrons. Anexample of a
calculation for a low earthorbit (LEO) (altitude of 500 km and latitude of 60 degrees) at solar
minimum and maximum is presented in Fig. 4 [1]. Solar minimum and maximum refer to
periods of minimum and maximum solar activity. Note that the flux of electrons decreases
rapidly athigh energies.
2.2 Cosmic Rays
Cosmic rays originate from two sources, the sun (solar) and sources outside our solar
system (galactic). Galactic cosmic rays are always present. In the absence of solar activity,
cosmic radiation is composed entirely of galactic radiation. Outside of our solar system, the
spectrum of galactic cosmic rays is believed to be uniform. Its composition as a function of
atomic mass is given in Fig. 5 [2,7]. It consists mostly of protons (85%) and alpha particles
(helium nuclei) (14%). Less than 1%of the galactic cosmic ray spectrum is composed of high-
energy heavy ions. This is not an indication that heavy ions arenot as important as protons in
II-6
...... "'_* '1 ' 'r' .... '/ +++
lOe ..... I ' " " I /
,,,J
10e -
___ 107
O--_l_ 1,0s
NO
Z_) 104
XO_ 10s
<C 10_
ILl 101
k.-
Z
-- 100
1 2 3 4 5 6 7
Earth Radii
Figure3: Distributionofprotonfluxasafunctionofenergyandradialdistance.(AfterRef.1)
spaceradiationeffects. As will be discussedbelow, heavy ions deposit more energyperunit
depthinamaterialthanprotons,andcan actually causegreaternumbersof single-eventeffects.
As illustratedinFig. 5,thefluxof protonsis morethantwo ordersof magnitudehigherthan the
fluxof eithercarbonoroxygenandapproximatelyfive ordersof magnitudehigherthanthe flux
of nickel. The energyspectrumof galactic cosmic rays is given in Fig. 6 [8]. Note that the
x-axisofFig. 6isgiveninunitsof MeV/nucleon. Thus,forcarbonwith 12nucleons,the pointat
100MeV/nucleonon the x-axis correspondsto an energyof 1.2GeV. Formost ions, the flux
peaks between 100 and 1000MeV/nucleon. For carbon, the peak flux is at an energy of
approximately2.4GeV. Forprotonsandalphaparticles,the energyof the ioncan bemorethan
100GeV/nucleon. At these highenergies, it is nearly impossibleto shieldelectronics inside a
spacecraftfromcosmicrays.
As cosmic rayspenetrate into the magnetosphere,low-energyparticles are attenuated,
modifyingthe cosmic rayspectrum. Onlythe moreenergeticparticles are ableto penetratethe
magnetosphere. Figure7 [1] illustratesthe attenuationof low-energyparticlesfor a low-earth
II-7
| I
"_ 1011
#.
_ 101o -
,,-I,lO,
"_ 10s
_C_Ul0 s !
lO,r DAILuYeT.o. FLUXBS
- lAD)
a 10s ORBIT:60°/500km
ILl
f<_ 102 r
IIC
10110-2 10-1 10o 101
< ENERGY (MeV)
Figure 4: Calculatedfluxofelectronsusing theAE8 modelfora.low-earthorbit. (AfterRef.1)
orbit (LEO) for several angles of inclination. Note that geomagnetic shielding decreases with
higher inclination orbits as the magnetic field lines converge nearthe poles.
The amount of solar cosmic raysis naturally dependent on the amount of solar activity.
Solar flares are random in nature and account for a large partof all solar cosmic rays. After a
solarflare occurs, particles begin to arrivenear the earth within tens of minutes, peakin intensity
within two hours to one day, and are gone within afew days to one week (except forsome solar
flare particles which are trappedin the earth's radiation belts). Ina solar flare, energetic protons,
alpha particles and heavy ions are emitted. Inmost solar flares the majority of emitted particles
are protons (90-95%) and alpha particles. Heavy ions constitute only a small fraction of the
emitted particles, and the number of heavy ions is normally insignificant compared to the
background concentration of heavy ions from galactic cosmic rays. In a large solar flare the
numberof protons and alpha particles can be greatly enhanced (~104times) over the background
galactic cosmic ray spectrum; whereas, the number of heavy ions for a large solar flare
approaches up to ~50% of the background galactic cosmic concentration of heavy ions [9].
Associated with a solar flare is the solar wind or solar plasma. The solar wind usually arrives
near the earth within one to two days after a solar flare [10]. As the solar wind strikes the
magnetosphere, it can cause disturbances in the geomagnetic fields (geomagnetic storm),
compressing them towards theearth. As aresult, the solarwind can enhance the total-dose that a
device receives inalow-earth orbit.
11-8
10.2 '' _1, I ,., .....i ........,,[, J , ,, , I,,, I , '_1[ i,,,,' ' , I ,_..',,,[,
0 10 20 30 40 50 60 70 80
ATOMICMASS
Figure 5: Fluxofgalacticcosmicrayparticlesforatomicmassesupto60. (AfterRefs.2and7)
Figure8 [11] is aplotof theangularflux of cosmicrayparticles(bothsolarandgalactic)
duringsolarminimumandmaximuminsideaspacecraftingeosynchronousorbitwith25milsof
aluminumshieldingas a functionof linearenergytransfer(LET). [LETis the mass stopping
powerof cosmic raysandis givenintheunitsof MeV/mg/cm2. Itis ameasureof theamountof
energyaparticletransferstoamaterialperunitpathlength.] Thesolarcycleisapproximately22
years long with peaks in intensity approximatelyevery 11 years. Solar maximum refersto
periodsof maximumsolar activity, and solar minimumrefers to periodsof minimumsolar
activity. Thesolar windduringperiodsof high solaractivityreducesthe galactic cosmic ray
flux. Thus,the minimum in galactic cosmic rayflux occursduringsolarmaximum, andthe
maximumingalacticcosmic rayfluxoccursduringsolarminimum.Thefluxatsolarmimmum
describesthe actualenvironmentfor40%of thetime. Also showninFig. 8 is theAdams'10%
worst-caseenvironment. Theactualenvironmentis moreintensethanthe Adams'10%worst-
case environmentonly 10%of the time. Itincludescontributionsfrombothgalacticandsolar
cosmic rays, This environmentis often used in assessing the single-eventupset hardnessof
electronicdevices.
2.3 RadiationEnvironmentInsideaSpacecraft
Thus far,we have exploredthe naturalspace radiationenvironmentoutside a spacecraft.
Todeterminethe effectsof the naturalspace environmentonelectronicsinsidethe spacecraft,the
effects of shielding must be taken into account. Shielding not only modifies the radiation
environmentinside a spacecraftby alteringthe energyand concentrationof incoming particles,
II-9
r_
!
102 -i-- i ...... i.............. I..............
100
He
1100-11 PROTONS -
1100_4 Fe FITO S
C&
X
111-7
101 10_ 104 104 10s 104
PARTICLE KINETIC ENERGY (MeV/nucleon)
Figure 6: Energy spectrum of galactic cosmic rays. (After Ref. 8)
but alsocancreatesecondaryparticlesasthe incomingparticlespassthroughthe shielding. For
instance, bremsstrahlungradiationin the form of x rays is emitted as energetic electrons
deceleratein the shielding. Formodestamountsof shielding, the effects of shieldingcanbe
estimated by takinginto accountonly the energy loss of particlesas they pass through the
shielding[11].
Theamountof energyloss as aparticlepassesthroughshieldingdependsonthethickness
of the material. Typicalspacecraftshieldingis intherangeof 100to250mils. Figure9a[12] is
aplotof fluxforalargesolarflareversusLETforaluminumthicknessesof 0.173to 10.8g/cm2.
Note thatincreasingaluminumthickness resultsin decreasingsolarflare fluxforthe relatively
low-energyparticlesassociatedwithasolar flare. However,thequalitativevariationinfluxwith
LETis relativelyunaffectedby the shielding. ForLETsabove30MeV-cm2/mgincreasingthe
shieldingthickness from0.17 g/cm2(25 mils) to 10.8g/cm2(1570 mils)reducesthe intensityof
the spectrumby five ordersof magnitude.Theeffect of spacecraftthicknesson galacticcosmic
rayfluxis showninFig.9b [12]. Ittakesmuchmoreshieldingto reducethe intensityof galactic
cosmic rays. Spacecraftthicknesses of aluminumfromzero up to 10g/cm2(1450 mils) only
slightly affect the LETspectrum. By comparingFigs. 9a and9b, we concludethat spacecraft
shielding can attenuatethe low-energynuclei from a solar flare, but has little effect on the
attenuationof nuclei in the galactic cosmic ray spectrum. Thus, for practical shielding
thicknesses, additionalshielding may proveeffective againstsoft componentsof a solarflare
environment,butisrelativelyineffectiveinreducingthegalacticcosmicrayspectrum[2].
H-iO